Results TIME models SUBSAMPLING plots 5ha

Data

Organising all tables of models results for all6 forest plo in one object per vital rate, quadrat size scale.

See table S1 for forest plot names.

# Pre-analysis
#Figures colors/labels
cores <- c( quadrat =  "#c00000", 
           `sp:time`= "#203864",
            time = "#ed7d31",
           `quadrat:time` = "#843c0c", 
           `quadrat:sp`= "#b666d2", 
            sp = "#70ad47", 
            Residual="#ffc000")

labvit <- as_labeller(c(grow = "Growth", mort="Mortality", 
                        rec="Recruitment"))
grpTlab = as_labeller(c(quadrat = "Space", `sp:time`="Species x Time", 
                        `quadrat:time`="Space x Time", `quadrat:sp`= "Species x Space",
                        sp="Species", Residual = "Residual",
                        time= "Time"))
grpvitT <- as_labeller(c(grow = "Growth", mort="Mortality", 
                         rec="Recruitment",
                        quadrat = "Space", `sp:time`="Species x Time", 
                         `quadrat:time`="Space x Time", `quadrat:sp`= "Species x Space",
                         sp="Species", Residual = "Residual",
                        time= "Time"))
labplot = as_labeller(c(quadrat = "Space", `sp:time`="Species x Time", 
                `quadrat:time`="Space x Time", `quadrat:sp`= "Species x Space",
                        sp="Species", Residual = "Residual",
                        time= "Time",
                fus="Fushan", luq = "Luquillo", lam="Lambir", 
                pas = "Pasoh", bci="Barro Colorado"))

Growth

Import results data to list

path = here("models_outputs/models_time/all", "grow", "table")
files = list.files(path)
 
all = lapply(files, function(x) get(assign(x, load(paste0(path, "/", x)))))

tudo <- bind_rows(all)

Remove intercept to include again as a separate column

intercepts <- tudo %>% filter(term == "intercept")
names(intercepts)[8] = "intercept"

growT <- tudo %>% filter(term != "intercept") %>%
  left_join(intercepts[,c(1:4,8)], by= c("sub","data", "fplot", "q_size")) %>%
  group_by(sub, data, fplot, q_size) %>% 
  mutate(VPC = variance/sum(variance))  %>% ungroup() %>%
  mutate(term = fct_relevel(term, "quadrat", "sp:time","time",
                            "quadrat:time",
                            "quadrat:sp", "sp", "Residual"),
         q_size = fct_recode(q_size, `5`= "quad_5", `10`= "quad_10",
                             `20`="quad_20",
                            `50`="quad_50",  `100`="quad_100")) %>%
  mutate(q_size = fct_relevel(q_size, "5","10", "20","50", "100")) %>%
  as.data.frame()

Divide the sd of the terms by the mean growth (intercept) for each dataset.

growT$stdev = growT$Estimate/growT$intercept

Number of growth models per florest plot and quadrat scale.

table(growT$fplot, growT$q_size)/7

##      
##        5 10 20 50 100
##   bci 10 10 10 10  10
##   fus 10 10 10 10  10
##   lam 10 10 10 10  10
##   luq 10 10 10 10  10
##   pas 10 10 10 10  10

Mortality

Import results data to list

path = here("models_outputs/models_time/all", "mort", "table")
files = list.files(path)
 
all = lapply(files, function(x) get(assign(x, load(paste0(path, "/", x)))))
tudo <- bind_rows(all)

Remove intercept to include again as a separate column

intercepts <- tudo %>% filter(term == "intercept")
names(intercepts)[8] = "intercept"

mortT <- tudo %>% filter(term != "intercept") %>%
  left_join(intercepts[,c(1:4,8)], by= c("sub","data", "fplot", "q_size")) %>%
  group_by(sub,data, fplot, q_size) %>% 
  mutate(VPC = variance/sum(variance))  %>% ungroup() %>%
   mutate(term = fct_relevel(term,"quadrat", "sp:time","time",  "quadrat:time",
                            "quadrat:sp", "sp", "Residual"),
         q_size = fct_recode(q_size, `5`= "quad_5", `10`= "quad_10",
                             `20`="quad_20",
                            `50`="quad_50",  `100`="quad_100")) %>%
  mutate(q_size = fct_relevel(q_size, "5","10", "20","50", "100")) %>%
  as.data.frame()

mortT$stdev <- mortT$Estimate

Number of mortality models per florest plot and quadrat scale.

table(mortT$fplot, mortT$q_size)/7

##      
##        5 10 20 50 100
##   bci 10 10 10 10  10
##   fus 10 10 10 10  10
##   lam 10 10 10 10  10
##   luq 10 10 10 10  10
##   pas 10 10 10 10  10

Recruitment

Import results data to list

path = here("models_outputs/models_time/all", "rec", "table")
files = list.files(path)
 
all = lapply(files, function(x) get(assign(x, load(paste0(path, "/", x)))))

tudo <- bind_rows(all)

Remove intercept to include again as a separate column

intercepts <- tudo %>% filter(term == "intercept")
names(intercepts)[8] = "intercept"

recT <- tudo %>% filter(term != "intercept") %>%
  left_join(intercepts[,c(1:4,8)], by= c("sub","data","fplot", "q_size")) %>%
  group_by(sub,data, fplot, q_size) %>% 
  mutate(VPC = variance/sum(variance))  %>% ungroup() %>%
   mutate(term = fct_relevel(term,"quadrat", "sp:time", "time", "quadrat:time",
                            "quadrat:sp", "sp", "Residual"),
         q_size = fct_recode(q_size, `5`= "quad_5", `10`= "quad_10",
                             `20`="quad_20", `50` = "quad_50",
                             `100`="quad_100")) %>%
  mutate(q_size = fct_relevel(q_size, "5","10", "20","50", "100")) %>%
  as.data.frame()

recT$stdev <- recT$Estimate

Number of recruitment models per florest plot and quadrat scale.

table(recT$fplot, recT$q_size)/7

##      
##        5 10 20 50 100
##   bci 10 10 10 10  10
##   fus 10 10 10 10  10
##   lam 10 10 10 10  10
##   luq 10 10 10 10  10
##   pas 10 10 10 10  10

combining rates

All vital rates

vitalT <- bind_rows(list(grow = growT, mort=mortT, rec=recT),.id="vital")

To get ONE value per forest-vital, first I calculate the mean of the 10 medians of the posterior variances. Then I use this value as the Variance and take the VPC and R2.

mvitalT <- vitalT %>% group_by(vital,data,fplot,q_size,term) %>%
  summarise(variance = mean(variance),
            stdev = mean(stdev)) %>%
   mutate(VPC = variance/sum(variance)) %>%
  mutate(term = fct_relevel(term, "quadrat", "sp:time","time", "quadrat:time", 
                            "quadrat:sp", "sp","Residual"))

mr2 <- mvitalT %>% filter(term != "Residual") %>%
  group_by(vital, data,fplot,q_size) %>% 
  summarise(r2 = sum(VPC))

Saving objects for models results

save(mvitalT, vitalT, file= here("models_outputs", "all_results_time.Rdata"))

All the 5ha subsampled plots

SDs

fsd <- vitalT %>% filter(q_size == "5") %>%
  ggplot(aes(x=fplot, y=stdev, col=term)) + 
  geom_quasirandom(alpha=0.5, groupOnX = TRUE) +
  facet_grid(vital~term) +
  stat_summary(fun=median, size=0.3, fatten=1,col="black",
                         geom="crossbar", width=0.7)+
  scale_color_manual(values = cores) +
  theme(legend.position = "none",
        panel.background = element_rect(colour="lightgray"),
        axis.text.x = element_text(angle=45,hjust=1)) +
  labs(title= "SD per subsample of 5ha - 5x5 quadrat scale")
fsd

vitalT %>% filter(vital=="grow") %>%
  ggplot(aes(x=q_size, y=stdev, col=term)) + 
  geom_quasirandom(alpha=0.5, groupOnX = TRUE) +
  facet_grid(fplot~term, labeller=labplot) +
  
  stat_summary(fun=median, size=0.3, fatten=1,col="black",
                         geom="crossbar", width=0.7)+
  scale_color_manual(values = cores) +
    xlab("Spatial scale") +
  theme(legend.position = "none",
        panel.background = element_rect(colour="lightgray"),
        axis.text.x = element_text(angle=45,hjust=1)) +
  labs(title= "SD per subsample of 5ha - GROWTH")

vitalT %>% filter(vital=="mort") %>%
  ggplot(aes(x=q_size, y=stdev, col=term)) + 
  geom_quasirandom(alpha=0.5, groupOnX = TRUE) +
  facet_grid(fplot~term, labeller=labplot) +
  
  stat_summary(fun=median, size=0.3, fatten=1,col="black",
                         geom="crossbar", width=0.7)+
  scale_color_manual(values = cores) +
    xlab("Spatial scale") +
  theme(legend.position = "none",
        panel.background = element_rect(colour="lightgray"),
        axis.text.x = element_text(angle=45,hjust=1)) +
  labs(title= "SD per subsample of 5ha - MORTALITY")

vitalT %>% filter(vital=="rec") %>%
  ggplot(aes(x=q_size, y=stdev, col=term)) + 
  geom_quasirandom(alpha=0.5, groupOnX = TRUE) +
  facet_grid(fplot~term, labeller=labplot) +
  
  stat_summary(fun=median, size=0.3, fatten=1,col="black",
                         geom="crossbar", width=0.7)+
  scale_color_manual(values = cores) +
    xlab("Spatial scale") +
  theme(legend.position = "none",
        panel.background = element_rect(colour="lightgray"),
        axis.text.x = element_text(angle=45,hjust=1)) +
  labs(title= "SD per subsample of 5ha - RECRUITMENT")

VPCs

fvpc <- vitalT %>% filter(q_size =="5") %>%
  ggplot(aes(x=fplot, y=VPC, col=term)) + 
  geom_quasirandom(alpha=0.5) +
  facet_grid(vital~term, labeller= grpvitT ) +
  stat_summary(fun=median, size=0.3, fatten=1,col="black",
                         geom="crossbar", width=0.7)+
  scale_color_manual(values = cores) +
  theme(legend.position = "none",
                panel.background = element_rect(colour="lightgray"),
        axis.text.x = element_text(angle=45,hjust=1)) +
  labs(title= "VPC per subsample of 5ha - 5x5")
fvpc

fig <- vitalT %>% filter(vital=="grow") %>%
  ggplot(aes(x=q_size, y=VPC, col=term)) + 
  geom_quasirandom(alpha=0.5, groupOnX = TRUE) +
  facet_grid(fplot~term, labeller=labplot) +
  
  stat_summary(fun=median, size=0.3, fatten=1,col="black",
                         geom="crossbar", width=0.7)+
  scale_color_manual(values = cores) +
    xlab("Spatial scale") +
  theme(legend.position = "none",
        panel.background = element_rect(colour="lightgray"),
        axis.text.x = element_text(angle=45,hjust=1)) +
  labs(title= "VPC per subsample of 5ha - GROWTH")
fig

fig <- vitalT %>% filter(vital=="mort") %>%
  ggplot(aes(x=q_size, y=VPC, col=term)) + 
  geom_quasirandom(alpha=0.5, groupOnX = TRUE) +
  facet_grid(fplot~term, labeller=labplot) +
  
  stat_summary(fun=median, size=0.3, fatten=1,col="black",
                         geom="crossbar", width=0.7)+
  scale_color_manual(values = cores) +
    xlab("Spatial scale") +
  theme(legend.position = "none",
        panel.background = element_rect(colour="lightgray"),
        axis.text.x = element_text(angle=45,hjust=1)) +
  labs(title= "VPC per subsample of 5ha - MORTALITY")
fig

fig <- vitalT %>% filter(vital=="rec") %>%
  ggplot(aes(x=q_size, y=VPC, col=term)) + 
  geom_quasirandom(alpha=0.5, groupOnX = TRUE) +
  facet_grid(fplot~term, labeller=labplot) +
  
  stat_summary(fun=median, size=0.3, fatten=1,col="black",
                         geom="crossbar", width=0.7)+
  scale_color_manual(values = cores) +
    xlab("Spatial scale") +
  theme(legend.position = "none",
        panel.background = element_rect(colour="lightgray"),
        axis.text.x = element_text(angle=45,hjust=1)) +
  labs(title= "VPC per subsample of 5ha - RECRUITMENT")
fig

Average per forest

SD stacked per forest

5x5 m

mvitalT %>% filter(q_size == "5") %>%
ggplot(aes(x=fplot, y=stdev, fill=term, col=term))+
  geom_col(alpha=0.7)+
  scale_color_manual(values = cores) +
  scale_fill_manual(values = cores)+
  facet_grid(~vital)

VPC per forest

5x5 m

ah <- mvitalT %>% filter(q_size == "5") %>% group_by(vital,term) %>%
  summarise(VPC_mean = as.character(round(mean(VPC), 2))) %>%
  mutate(x_val = as.numeric(VPC_mean), 
         y_val = 0.6:6.8)
ah[c(11,13),3] <-  c("0.10", "0.30")

pm<- mvitalT %>% filter(q_size == "5") %>%
  ggplot(aes(y=rev(term), x=VPC)) +
  facet_grid(~vital, labeller = grpvitT, scales="free_x") +
  geom_jitter(size=5, alpha=0.5, aes(shape=fplot, fill=term), width = 0,
              height = 0.15, col="black") +
  stat_summary(fun=mean, fun.min=mean, fun.max=mean,
               geom="crossbar", width=0.6, fatten=3) +
  scale_fill_manual(name= "Forest", values=cores, guide ="none")+
  scale_shape_manual(name="Forest",values=c(22,21,24,23,25), #c(15,16,17,18,8)
                     labels= c("Barro Colorado", "Fushan", "Lambir","Luquillo",
                               "Pasoh", "Wytham Woods")) +
  geom_vline(xintercept=c(0,0.2,0.4,0.6), col='gray', linetype="dashed",
             size=0.2) +
  scale_y_discrete(labels=c("Residual", "Species", "Species x Space", 
                                "Space x Time", "Time", "Species x Time",
                                "Space"))+
  theme(legend.position= "bottom",
        plot.margin =  margin(0.1, 0.1, 0.1, 0, "cm"),
        legend.background = element_rect(fill="gray95"),
        panel.spacing.y = unit(0.8, "cm"),
        panel.background = element_rect(colour="lightgray"),
        text = element_text(size=16),
        axis.text = element_text(size=14)) +
  ylab("") + xlab("VPC") +
  geom_text(data=ah, aes(x=x_val, y=rev(y_val), label=VPC_mean), hjust=0,
            size=4)
pm

mvitalT.sum <-  mvitalT %>% filter(q_size=="5") %>% ungroup()%>%
  group_by(vital,term) %>%
  summarise(meanVPC = mean(VPC),
            sdVPC = sd(VPC)) %>%
  pivot_wider(names_from = vital, values_from = c(meanVPC, sdVPC)) %>%
  mutate_if(is.numeric, round, digits=2) %>%
  select(term, meanVPC_grow, sdVPC_grow, meanVPC_mort, sdVPC_mort,
         meanVPC_rec, sdVPC_rec)


mvitalT.sum %>%
  htmlTable(header=c("term" ,rep(c("Mean", "SD"), 3)),
            cgroup = c("", "Growth", "Mortality", "Recruitment"),
            n.cgroup =c(1,2,2,2))

		Growth		Mortality		Recruitment
	term	Mean	SD	Mean	SD	Mean	SD
1	quadrat	0.02	0.02	0.02	0.02	0.04	0.04
2	sp:time	0.02	0.02	0.02	0.01	0.04	0.01
3	time	0.03	0.01	0.06	0.05	0.09	0.04
4	quadrat:time	0.03	0.02	0.1	0.07	0.15	0.07
5	quadrat:sp	0.25	0.07	0.06	0.02	0.1	0.07
6	sp	0.19	0.08	0.26	0.15	0.22	0.11
7	Residual	0.48	0.07	0.48	0.06	0.36	0.09

TOTAL VPC per vital rate 5x5

To make the bar plots, first I sum all the variances per term and then calculate VPC from this sum

tot.vpc.fig <- mvitalT %>% filter(q_size=="5") %>% ungroup()%>%
  group_by(vital,term) %>%
  summarise(sumVar =sum(variance)) %>%
  mutate(VPC = sumVar/sum(sumVar)) %>%
ggplot(aes(x=vital, fill=term, col=term, y=VPC)) + 
  geom_col(alpha=0.7, width = 0.6) +
  facet_grid(~vital, scales="free_x", labeller = labvit) +
   xlab("") + ylab("Mean VPC") +
    scale_fill_manual(values=cores, labels=grpTlab) +
  scale_color_manual(values=cores, labels=grpTlab) +
  theme(legend.position = "none",
         panel.spacing.x = unit(0.01, "cm"),
        axis.text.x = element_blank(),
        axis.ticks = element_blank(),
            text = element_text(size=16),
        axis.text = element_text(size=14))
tot.vpc.fig

FIGURE 2

pm +  labs(tag="(a)") +
  theme( plot.margin =  margin(0.1, 0.1, 0.1, 0, "cm"),
         plot.tag.position = c(0,0.93),
         plot.tag = element_text(size=20, vjust = -0.6, hjust = -1)) +
tot.vpc.fig + labs(tag="(b)") +
    theme(plot.tag.position = c(0,0.93),
         plot.tag = element_text(size=20, vjust = -0.6, hjust = 0.1)) +
    plot_layout(widths = c(1,0.4))

ggsave(here("figures","FIG_2.jpeg"), width=14, height=6)

FIGURE 3 VCP bar all scales

mvitalT %>% group_by(vital,q_size,term) %>%
  summarise(sumVar =sum(variance)) %>%
  group_by(vital,q_size)%>%
  mutate(VPC = sumVar/sum(sumVar)) %>% select(-sumVar) %>%
  pivot_wider(names_from =q_size, values_from = VPC) %>%
  mutate_if(is.numeric, round, digits=2) %>%
  htmlTable()

	vital	term	5	10	20	50	100
1	grow	quadrat	0.01	0.02	0.01	0.01	0.01
2	grow	sp:time	0.02	0.02	0.02	0.02	0.01
3	grow	time	0.03	0.03	0.02	0.02	0.02
4	grow	quadrat:time	0.02	0.02	0.01	0.01	0
5	grow	quadrat:sp	0.25	0.2	0.14	0.08	0.04
6	grow	sp	0.2	0.19	0.19	0.2	0.2
7	grow	Residual	0.47	0.54	0.61	0.67	0.71
8	mort	quadrat	0.02	0.01	0.01	0.01	0.01
9	mort	sp:time	0.02	0.02	0.02	0.02	0.02
10	mort	time	0.06	0.06	0.06	0.06	0.05
11	mort	quadrat:time	0.09	0.06	0.03	0.01	0.01
12	mort	quadrat:sp	0.06	0.06	0.05	0.03	0.02
13	mort	sp	0.27	0.28	0.27	0.28	0.28
14	mort	Residual	0.48	0.51	0.55	0.58	0.6
15	rec	quadrat	0.04	0.03	0.02	0.01	0.03
16	rec	sp:time	0.04	0.04	0.04	0.04	0.05
17	rec	time	0.09	0.1	0.11	0.12	0.1
18	rec	quadrat:time	0.15	0.13	0.11	0.08	0.1
19	rec	quadrat:sp	0.11	0.1	0.09	0.06	0.05
20	rec	sp	0.22	0.24	0.24	0.24	0.24
21	rec	Residual	0.34	0.37	0.39	0.43	0.43

barvpc <- mvitalT %>% group_by(vital,q_size,term) %>%
  summarise(sumVar =sum(variance)) %>%
  group_by(vital,q_size)%>%
  mutate(VPC = sumVar/sum(sumVar)) %>%
  ggplot(aes(x=q_size, y=VPC, fill=term, col=term))+
  geom_col(alpha=0.7)+
  scale_color_manual(values = cores, labels=grpTlab) +
  scale_fill_manual(values = cores, labels=grpTlab )+
  facet_grid(~vital,labeller = grpvitT) +
  xlab("Spatial grain (quadrat size)") +
  ylab("VPC") + labs(tag="(b)")
barvpc

png(here("figures", "FIG_3_VPC_time_scales.png"), height = 550, width=1300, res=130)
barvpc
dev.off()

## quartz_off_screen 
##                 2